Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

"So, you're *my* brother?" Mark's voice dripped with sarcasm. He was leaned back against the peeling paint of the rental house's front porch, arms crossed, as if to shield himself. Across from him, Daniel, the supposed sibling, fidgeted on the cracked concrete steps, his nervous energy a stark contrast.

The lawyer had explained the situation, the DNA tests, the shared inheritance. Mark hadn't cared about the money. He cared about the years of his life, spent believing he was an only child, then this… this stranger. He ran a hand through his hair, the gesture abrupt.

Daniel’s story, a tale of a long-lost father who never knew, sounded rehearsed. Mark wasn't buying it. He saw only a desperate grab for a piece of the pie. A sudden, potent dislike settled in his gut.

He smirked. "Well, Daniel, I guess we’re family now. How *delightful*."

Emotion: contemptuous

Cluster: Anger / Hostility
PC1 (Valence): -0.40 Negative
PC2 (Disposition): 2.07

Role in Research

This story is one of 1,000 stories generated for the emotion contemptuous. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 contemptuous stories, after denoising with neutral dialogue baselines, produces the contemptuous emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the contemptuous vector is projected through the unembedding matrix.

Promoted:
C0.502
l0.423
fact0.287
either0.265
👎0.262
Suppressed:
own-0.367
unexpectedly-0.303
serendip-0.265
Suddenly-0.253
previously-0.250